Abstract:Automatic Speech Recognition (ASR) can significantly reduce documentation burden in clinical workflows, but standard models degrade sharply in real-world telephony settings where noisy audio, dialectal variation, and strict data residency constraints prevent cloud-based adaptation. We study this "reality gap" using Gram Vaani: a telephonic Hindi corpus spanning rural healthcare and agricultural helplines, as the closest available proxy for clinical speech under strict on-device constraints. We show that a robust multilingual model (IndicWav2Vec) degrades from 11.59\% WER on standard clean Hindi to \textbf{41.71\% WER} on this proxy telephony data. We evaluate a progression of on-device adaptation regimes under realistic constraints, from full fine-tuning to parameter-efficient LoRA and stream-based continual learning, across multiple baselines, datasets, and seeds. Focusing on continual learning, our central finding highlights a critical interaction between Experience Replay (ER) and Elastic Weight Consolidation (EWC, parameterized by regularization strength $\lambda$). We show that standard positive EWC ($\lambda > 0$) can oppose replay-driven updates, limiting adaptation. Reversing EWC's strength ($\lambda < 0$) suggests that it can act as a directional control signal under ER-guided adaptation: negative $\lambda$ reinforces replay-driven plasticity, while a scheduled $\lambda$ enables phase-dependent control of stability and plasticity. Across evaluations on multiple datasets, we find that multi-domain replay provides a strong foundation for adaptation, while EWC modulates stability-plasticity dynamics without altering final performance. These results show that effective on-device adaptation depends on understanding how data-driven and parameter-level learning signals interact, rather than choosing methods in isolation.
| Comments: | 17 pages. Under review |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2512.16401 [cs.CL] |
| (or arXiv:2512.16401v5 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2512.16401 arXiv-issued DOI via DataCite |
Submission history
From: Darshil Chauhan [view email]
[v1]
Thu, 18 Dec 2025 10:56:27 UTC (5,392 KB)
[v2]
Mon, 22 Dec 2025 16:22:23 UTC (5,941 KB)
[v3]
Thu, 1 Jan 2026 16:03:15 UTC (5,944 KB)
[v4]
Wed, 14 Jan 2026 15:22:47 UTC (5,949 KB)
[v5]
Sun, 31 May 2026 17:01:07 UTC (5,952 KB)